Rapid Characterization of Chemically-Modified Proteins by

Sep 15, 1995 - Australia, and Biomedicine and Health Program, Australian Nuclear Science and ... Lucas Heights, New South Wales 2234, Australia...
0 downloads 0 Views 323KB Size
Bioconjugate Chem. 1996, 7, 16−22

16

Rapid Characterization of Chemically-Modified Proteins by Electrospray Mass Spectrometry Keiryn L. Bennett,†,‡ Suzanne V. Smith,‡ Richard M. Lambrecht,† Roger J. W. Truscott,† and Margaret M. Sheil*,† Department of Chemistry, University of Wollongong, Northfields Avenue, Wollongong, New South Wales 2522, Australia, and Biomedicine and Health Program, Australian Nuclear Science and Technology Organisation, Lucas Heights, New South Wales 2234, Australia. Received March 24, 1995X

Electrospray mass spectrometry (ESI-MS) has been used to examine monoclonal antibodies (MAbs), antibody fragments (Fab and Fc), modified fragments, and a range of other chemically-modified proteins as part of a study aimed at establishing ESI-MS as a method for the characterization of radioimmunoconjugates. This has been approached from two angles. Firstly, ESI-MS of complexes formed between chelators and other small molecules conjugated to hen egg white lysozyme (HEL) (14 kDa) demonstrate the considerable advantages of this powerful new technique compared with existing methods for the characterization of chemically-conjugated proteins. Molecular weights can be determined rapidly to within 0.01-0.05% and with good sensitivity (10-50 pmol total), thus providing specific structural information and opening the way for ESI-MS to be applied widely for the structural characterization of radioimmunoconjugates. Secondly, the conditions for ESI-MS of intact antibodies and antibody fragments have been examined in detail, and we have shown that the addition of up to 10 biotin molecules to the 50 kDa Fab fragment can be easily detected in ESI mass spectra, thus demonstrating the potential for the characterization of modified MAb fragments and metabolites. Finally, the strengths and limitations of ESI-MS of intact antibodies are discussed, and these results indicate that it may only be possible to detect average shifts in the mass of intact antibodies following modification.

INTRODUCTION

Radioimmunospecific pharmaceuticals, in which radionuclides are incorporated into a monoclonal antibody (MAb) or genetically engineered proteins, are contemporary approaches to diagnostic and therapeutic treatments of cancer (1, 2). Methods for labeling antibodies include the following: (i) direct labeling for isotopes such as iodine-123, iodine-131 (3), and technetium-99m (4) and (ii) indirect labeling with the use of bifunctional chelators for isotopes including indium-111, yttrium-90 (3), and rhenium-188 (4). Many of the factors limiting clinical applications of radioimmunoconjugates, for example, problems with the human anti-mouse antibody (HAMA) response, have been overcome by the use of antibody fragments and the development of recombinant forms of MAbs (5). However, the chemistry involved in modification of MAbs remains a significant limiting factor in the development of radioimmunoconjugates for clinical use. For the ongoing development of radioimmunoconjugates, it is vital to establish loading values of chelators on the proteins (i.e., average numbers of moles of chelator bound to the protein), to characterize the nature of the ligand binding, and to have a method for determining the purity of the resulting radiopharmaceutical. Current techniques available for assessment of loading conditions of MAbs are most commonly based on UV spectroscopy (6) and radioactivity trace assays (7). The accuracy of the UV spectroscopy assays depends on the relative absorptivities of the conjugate and the protein. Radioassays measuring radioisotope bound in an anti* To whom correspondence should be addressed. Phone: 6142-213261. Fax: 61-42-214287. E-mail: [email protected]. † University of Wollongong. ‡ Australian Nuclear Science and Technology Organisation. X Abstract published in Advance ACS Abstracts, September 15, 1995.

1043-1802/96/2907-0016$12.00/0

body-ligand conjugate may be subject to error if there are noncovalent interactions of the isotopes with the protein. Further, both UV and radioassays only produce an average measurement of ligand binding and yield no information concerning the sites and nature of the binding. Hillenkamp et al. (8) have shown matrixassisted UV laser desorption ionization (MALDI) mass spectrometry to be a useful tool for the determination of loading values of chelators bound to proteins. With this technique, average molecular masses of proteins up to 200 kDa can be determined using only very small amounts of protein (1-10 pmol). A complementary technique to MALDI-MS is electrospray mass spectrometry (ESI-MS) (9). The utility of ESI-MS for the characterization of proteins and large biomolecules is now well established (9-11). ESI-MS offers higher resolution and mass accuracy than MALDI. Ionization of the protein is achieved by evaporation of the solution under the influence of an electric field to yield a series of multiplycharged protonated molecular ions. For proteins, an envelope of multiply charged ions is normally observed with mass-to-charge ratios (m/z) within the range of 5004000. The ionization process is very gentle so little or no fragmentation of the large molecules occurs. While ESI-MS of monoclonal antibodies, antibody fragments, and recombinantly engineered antibodies have been published (8, 11-18), there have been no mass spectra reported for chemically-modified antibodies. We report here the development of techniques using ESI-MS for the characterization of coupled chelators and other labels using hen egg white lysozyme (HEL) (14 kDa) as a model to demonstrate the potential of ESI-MS for assessment of chemical protein conjugates. In addition, we have optimized conditions for ESI-MS of the murine B72.3 MAb (19), Fab, and Fc fragments, and we show the ESI-MS of the biotinylated B72.3 Fab fragment. Finally, © 1996 American Chemical Society

Bioconjugate Chem., Vol. 7, No. 1, 1996 17

Electrospray MS of Chemically-Modified Proteins

the strengths and limitations of ESI-MS for characterization of MAbs are discussed. EXPERIMENTAL PROCEDURES

Materials. Hen egg white lysozyme (HEL), poly(propylene glycol) 3000, anhydrous dimethyl sulfoxide (DMSO), 1-(3-(dimethylamino)propyl)-3-ethylcarbodiimide hydrochloride (EDC), mercuripapain (2 × crystallized suspension in 70% ethanol) were purchased from Sigma Pty. Ltd. (St. Louis, MO), Iodogen and (N-hydroxysuccinimido)biotin (NHS-biotin) were purchased from Pierce (Rockford, IL), and murine B72.3 monoclonal antibody was purchased from Bioquest Limited (Sydney, NSW, Australia). [Co(diAMsarH2)](NO3)2 (diAMsar ) 1,8-diamino-3,6,10,13,16,19-hezaazabicyclo[6.6.6]icosane) was a gift from Dr. Stephen Ralph, Department of Chemistry, University of Wollongong, Australia. Chemical Modification Reactions. Biotinylation. HEL or B72.3 Fab fragment (1 mg/mL) in 50 mM sodium bicarbonate buffer (pH 8.5) was reacted at 2- 5- and 8-fold molar ratios with NHS-biotin dissolved in anhydrous DMSO. Samples were incubated at room temperature for 30 min (20). Iodination. One hundred µL of Iodogen (0.5 mg/mL) in chloroform was pipetted into a test tube and the solvent evaporated to coat the tube with the oxidative agent. One hundred µL of HEL (5 mg/mL) in 10 mM sodium phosphate-buffered saline (pH 7.2) was pipetted into coated tubes, and varying molar ratios of Na127I were added. Samples were reacted at room temperature for 10 min before the reaction was quenched by decanting the mixture from the tube (21). Conjugation of Chelators. A 100-fold molar excess of EDC and a 1000-fold molar excess of [Co(diAMsarH2)](NO3)2 was added to HEL (10 mg/mL) in 75 mM phosphate buffer (final pH 4.75) and incubated at 37 °C for 60 min. Excess EDC and [Co(diAMsarH2)](NO3)2 were removed by ultrafiltration. This procedure was based on the method of Conrad et al. (22), although a 10-fold molar excess of EDC was used to conjugate Co(diAMsar)3+ (23) to cytochrome c. Samples were incubated at room temperature for 30 min. Preparation of Fab and Fc Fragments. Papain (4 mg/ mL) was diluted in digestion buffer (75 mM NaCl, 2 mM disodium EDTA, and 75 mM sodium phosphate, pH 7.0) and activated for 1-2 min at 37 °C before addition to the MAb. Fragments were prepared by dialyzing B72.3 MAb (10 mg/mL) into the digestion buffer containing 10 mM cysteine-HCl and then digesting with 10% papain (w/w) for 90 min at 37 °C. Separation of Fab and Fc Fragments. MAb fragments were separated on a FPLC system (Pharmacia, Uppsala, Sweden) by HR 10/10 Mono Q anion exchange with a linear gradient of 5-300 mM sodium phosphate buffer (pH 8) at a flow rate of 2 mL/min. The gradient was generated over 15 min. Conjugated protein samples, i.e., biotinylated HEL or Fab fragment, and iodinated and Co(diAMsar)-modified HEL were analyzed by ESI-MS without further purification. This is a key feature of the method enabling analysis of reaction mixtures to determine the individual species and their relative abundance. All samples were extensively dialyzed against water or desalted using Centricon-10 and Centricon-30 ultrafiltration tubes (Amicon, Lexington, MA). It is essential for ESI-MS analyses that stabilizing buffer salts and unreacted species are removed prior to analysis. The presence of salt in the sample both lowers ESI ionization efficiency and can reduce the accuracy of the mass measurement. Further, desalting reduces the tendency for excess unreacted materials to form nonspecific non-

covalent adducts often observed by ESI-MS (more below). Dialyzed material was lyophilized in a Speed Vac (Savant, Farmingdale, NY) and stored at -20 °C until required. Mass Spectrometry. A VG Quattro triple quadrupole mass spectrometer (Fisons Biotech MS, Altrincham, U.K.) was used for all experiments. Protein samples were dissolved in 50% aqueous acetonitrile plus 1% formic acid (pH 2.5). The samples were delivered to the instrument in 50% aqueous methanol via an ISCO SFC500 syringe pump (Lincoln, NE) at a flow rate of 5-10 µL/min. Ten µL of sample was injected for each analysis. HEL and Fc fragments were dissolved at a concentration of 20 pmol/µL, the Fab fragment at 10 pmol/µL, and the MAb required a minimum of 1 pmol/µL for optimum sensitivity. Similarly, the potential on the first source skimmer also required tuning for each protein. The skimmer potential for the MAb was 150 V, and 70-75 V was used for the other proteins. A dry nitrogen bath gas at atmospheric pressure was employed to assist evaporation of the solvent. The electrospray probe tip potential was 3.5 kV with 0.5 kV on the chicane counter electrode. The mass range was calibrated with HEL or ammoniated poly(propylene glycol) 3000 depending on the m/z range of the protein sample to be analyzed. The m/z scale in the raw spectra were converted to exact molecular weights using the transform function of the MassLynx software (Cambridge, U.K.). The m/z values for individual peaks are determined at the center of each peak at 80% height following smoothing. The uncertainties in the masses reported here are calculated from the range of masses generated by this function. For the modified Fab fragment, molecular weights and uncertainties were determined by deconvoluting the raw spectra using the more sophisticated maximum entropy (MaxEnt) method (24). RESULTS

Chemically-Modified Proteins. A range of different chemical labeling agents using a small protein as a model have been examined. Figure 1a-c shows the ESI mass spectra of HEL following biotinylation with (N-hydroxysuccinimido)biotin (NHS-biotin) in protein to biotin ratios of 1:2 (Figure 1a), 1:5 (Figure 1b), and 1:8 (Figure 1c). In each case the most abundant ions are due to the 9+ charge state, and the observed charge states range from 5+ to 10+. This compares (or contrasts) with the charge states observed for native HEL under the same solution conditions. Figure 2a-c shows the spectra corresponding to Figure 1a-c following transformation from a m/z to a mass scale. The addition of one biotin molecule to a protein increases its overall mass by 226 Da, and this is readily detected in the ESI mass spectra. Figure 2a clearly shows a peak due to the native HEL (14 306 ( 2 Da) and a peak due to the addition of 1 biotin (14 532 ( 2 Da) which is the most abundant species under these conditions. Peaks corresponding to the addition of two biotins (14 758 ( 2 Da), three biotins (14 986 ( 1 Da), four biotins (15 211 ( 3 Da), and five biotins (15 438 ( 5 Da) are also evident in this spectrum. As the ratio of biotin to protein was increased the distribution of molecular species shifted to higher masses. With a 1:5 HEL to biotin ratio, the most abundant species is HEL plus two biotins (Figure 2b), and with a 1:8 ratio, HEL plus four biotins is the most abundant (Figure 2c). In this case there is also a peak due to the addition of six biotins (15 665 ( 3 Da), and the native protein is no longer evident. There is no indication of more than six biotin molecules being incorporated into HEL even with the 1:8 ratio (Figure 2c). The molar incorporation of biotin into

18 Bioconjugate Chem., Vol. 7, No. 1, 1996

Figure 1. Electrospray mass spectra of biotinylated hen egg white lysozyme (HEL), skimmer potential 70 V, protein concentration 20 pmol/µL. Molar ratio of protein to biotin: (a) 1:2, (b) 1:5, (c) 1:8. Key: 1, HEL; (, HEL + 1 biotin; b, HEL + 2 biotin; *, HEL + 3 biotin; 2, HEL + 4 biotin; †, HEL + 5 biotin; 9, HEL + 6 biotin.

Figure 2. Electrospray mass spectra of biotinylated HEL transformed to a mass scale, skimmer potential 70 V, protein concentration, 20 pmol/µL. Molar ratio of protein to biotin: (a) 1:2, (b) 1:5, (c) 1:8.

HEL can be estimated by dividing the relative intensity of each peak in the transformed mass spectrum by the total intensity of the peaks in the spectrum, assuming the ionization efficiency for each species is comparable (25). This approach indicates that when the ratio of HEL to biotin was 1:2, 1.3 molecules of biotin were incorporated per molecule of HEL. This increased to 1.8 and 3.7 for the ratios of 1:5 and 1:8, respectively. Note that the small peaks approximately 50 Da higher than the HEL were also present in the spectrum of the native HEL (not shown) and are thus probably due to nonspecific salt adducts rather than a product arising from reaction with NHS-biotin.

Bennett et al.

The addition of radioisotopes of iodine to proteins has been well established as a method for direct labeling (26). ESI-MS spectra (not shown) were obtained following the reaction of HEL with mole ratios of protein to Na127I ranging from 1:1 to 1:100. These data, summarized in Table 1, demonstrate that even with a 100-fold molar excess of Na127I there is no evidence of more than four iodine-127 atoms being conjugated to HEL. Figure 3a shows the molar incorporation of iodine-127 versus molar ratio of Na127I added to HEL. These values are calculated from the relative intensities of iodinated HEL in the ESI spectra. The degree of incorporation initially increases logarithmically with increasing Na127I up to 1:10 and thereafter levels off to a average incorporation of 2.5 iodine-127 atoms per molecule of HEL. Figure 3b shows the percentage yield of iodinated HEL as a function of moles of Na127I added to the protein. These data indicate that at a molar ratio of protein to Na127I of 1:50 or greater there is a 96.3% to 98.1% incorporation of iodine-127 into the protein, i.e., the majority of the protein has been modified by iodine-127. A second method for attachment of a radioisotope or metal to a protein involves the use of a bifunctional chelator or bioconjugate, i.e., indirect labeling. There is increasing interest in using stable macrocyclic cage complexes in nuclear medicine, as retention of the radioisotope or metal within the cage structure is markedly improved in vivo (27). Thus, in the present study a readily available cage compound was used to demonstrate the utility of ESI-MS to examine protein conjugates of this type. Covalent attachment of the macrocyclic cage complex [Co(diAMsar)]3+ (23) to acidic protein residues (glutamic and aspartic acid) of HEL is achieved via 1-(3(dimethylamino)propyl)-3-ethylcarbodiimide hydrochloride (EDC) coupling. The ESI spectrum of HEL coupled to [Co(diAMsar)]3+ via EDC is shown in Figure 4. This shows the addition of one [Co(diAMsar)]3+ moiety to HEL at 14 662 ( 3 Da (theoretical mass 14 661 Da). There are also peaks corresponding to the N-acylurea protein adduct at 14 467 Da and to the EDC modified Co(diAMsar)-HEL conjugate at 14 818 Da. Analysis of B72.3 Monoclonal Antibody and Antibody Fragments. Illustrated in Figure 5 is the ESI mass spectrum for the intact murine B72.3 monoclonal antibody showing charge states 54+ to 39+. The molecular mass determined for this MAb was 146 861 ( 52 Da, which is approximately 3% higher than the mass of 143 756 Da calculated from the DNA sequence (28, 29). This is consistent with the known degree of glycosylation of the IgG class of MAb (30). Some of the experimental parameters required for observation of these large proteins by ESI differed to those for proteins of lower mass. These include the following: the need for a higher skimmer potential (150 V compared with 75 V for antibody fragments), a sharper dependence on protein concentration with an optimum range of 1-5 pmol/µL compared with 10-20 pmol/µL for smaller proteins, and a generally higher observed m/z range, being centered around m/z 3000 compared with m/z 1500 for HEL (Figure 1) and the reduced Fc fragment (Figure 6). Further details of B72.3 MAb ESI-MS optimization are reported elsewhere (31). The data processing is also important for these large proteins. The heterogeneity arising from glycosylation (32) causes broadening of the peaks in the ESI mass spectrum of the intact MAb and can result in large uncertainties in mass measurement. There may be also additional peak broadening associated with noncovalent salt adducts (31). The data in the spectrum shown in Figure 5 have been “smoothed” to enable an average

Bioconjugate Chem., Vol. 7, No. 1, 1996 19

Electrospray MS of Chemically-Modified Proteins

Table 1. Expected and Observed Masses for Iodinated Lysozyme at Varying Molar Ratios of Protein to Sodium Iodide (Na127I) observed mass (Da) component

expected mass (Da)

1:1

1:2

1:5

1:10

1:50

1:75

1:100

lysozyme + iodine-127 + 2 iodine-127 + 3 iodine-127 + 4 iodine-127 + 5 iodine-127 + 6 iodine-127

14 306 14 432 14 558 14 684 14 810 14 936 15 062

14 306 ( 1 14 431 ( 1 14 558 ( 4

14 307 ( 1 14 433 ( 2 14 559 ( 3

14 306 ( 2 14 434 ( 2 14 559 ( 2 14 687 ( 7 14 814 ( 6

14 308 ( 1 14 433 ( 1 14 559 ( 2 14 686 ( 2 14 813 ( 1

14 306 ( 3 14 434 ( 1 14 559 ( 1 14 684 ( 1 14 811 ( 1

14 307 ( 3 14 434 ( 1 14 559 ( 1 14 685 ( 1 14 811 ( 2

14 306 ( 2 14 432 ( 1 14 555 ( 2 14 683 ( 2 14 810 ( 2

Figure 4. Electrospray mass spectrum of HEL-[Co(diAMsar)] conjugate transformed to a mass scale, skimmer potential 70 V, protein concentration, 20 pmol/µL.

Figure 5. Electrospray mass spectrum of intact murine B72.3 monoclonal antibody, skimmer potential 150 V, protein concentration 1 pmol/µL.

Figure 3. (a) Molar incorporation of iodine into HEL versus molar ratio Na127I added. (b) Percentage yield of iodinated hen egg white lysozyme versus molar ratio Na127I added.

molecular weight to be determined from the data to overcome this problem. It was not possible to derive a molecular weight from unsmoothed data. It should also be noted that the peaks become more asymmetric as the charge state decreases. This has been attributed to nonspecific salt adducts which are present even after extensive dialysis (31). These are more pronounced at higher m/z as the salt adducts appear to block accessible charge states. Thus, we have based the mass calculation on the peak centers at 80% of the peak height to minimize the effect of this asymmetry. Furthermore, only the seven most intense charge states, i.e., 49+ to 43+, were included in the calculation. The same data processing

parameters for smaller proteins yield molecular masses to within 0.05%. The ESI spectrum of the reduced B72.3 Fc subunit is shown in Figure 6a. Charge states 14+ to 22+ are given, with an expanded view of charge state 19+ shown inset. The transformed ESI-MS spectrum shown in Figure 6b indicates the presence of three major components labeled A, B, and C. The predominant peak in the spectrum (component A) has an average molecular mass of 25 733 ( 7 Da, which is consistent with the amino acid sequence of this fragment (heavy chain fragment: Val-225 f Lys438) (28) plus the core carbohydrate portion (2 Nacetylglucosamine, 3 mannose, 1 fucose) with the addition of two N-acetylglucosamine units (more below). The other major peaks at 25 895 ( 6 Da (component B) and 26 057 ( 8 Da (component C) correspond closely to the addition of 1 and 2 galactose units, respectively. The ESI mass spectrum of the B72.3 Fab subunit showing charge states 23+ to 37+ is illustrated in Figure 7. The average molecular weight calculated from the raw

20 Bioconjugate Chem., Vol. 7, No. 1, 1996

Bennett et al.

Figure 8. Electrospray mass spectra of biotinylated Fab fragment converted to a mass scale using MaxEnt deconvolution software, skimmer potential 75 V, protein concentration 10 pmol/µL. Molar ratio of protein to biotin: (a) 1:2, (b) 1:5, (c) 1:8.

Figure 6. Reduced Fc fragment of the B72.3 monoclonal antibody, skimmer potential 75 V, protein concentration 20 pmol/µL. (a) Electrospray mass spectrum. The inset shows the expanded view of 19+ charge state. (b) Electrospray mass spectrum transformed to a mass scale. Glycosylation patterns are as indicated.

Figure 7. Electrospray mass spectrum of Fab fragment of the B72.3 monoclonal antibody, skimmer potential 75 V, protein concentration 10 pmol/µL.

data was 46 525 ( 8 Da. On the basis of the DNA sequence of the B72.3 MAb, the expected mass of the Fab fragment is 46 832 Da (light chain: Asp-1 f Cys-214 plus heavy chain fragment: Gln-1 f Gly-217) (28, 29). Thus, the lower mass observed by ESI-MS may suggest some additional enzymatic digestion or degradation. The ESIMS transformed spectra of biotinylated Fab fragment at ratios of 1:2, 1:5, and 1:8 are shown in Figure 8a-c. Successive increases in mass of only 226 Da on the 50 kDa protein fragment are clearly evident in the spectra. An average of 2.1 biotin molecules were incorporated into the B72.3 Fab fragment when the ratio of protein to biotin

was 1:2. This increased to 4.2 and 7.0 for the ratios of 1:5 and 1:8, respectively. It is interesting to note that even at a Fab to biotin ratio of 1:2, the native protein is no longer evident. This suggests a 100% molar incorporation of biotin since there is no purification stage between chemical modification and analysis. The data from ESI-MS of the B72.3 monoclonal antibody and fragments are summarized in Table 2. DISCUSSION

HEL was chosen as the initial model protein for this study as there is substantial literature available concerning the accessibility of the amino acid residues in this protein (33). (N-Hydroxysuccinimido)biotin (NHS-biotin) reacts specifically with primary amines (generally lysine  groups) on proteins to form stable amide bonds. HEL has six lysine residues available for conjugation to a biotin molecule, and as seen in Figure 2c all six residues appear to have been conjugated when a 8 mol excess of biotin was used. The possibility of reaction of NHS-biotin with the other amino groups on the protein (e.g., Nterminal) cannot be excluded on the basis of this data; however, there is no spectral evidence for the incorporation of a seventh biotin molecule. The available methods for attachment of iodine to proteins employ the use of oxidative agents for producing an electrophilic iodine species (I+), thereby enabling iodine to react with aromatic groups, with tyrosine residues the major site of iodination (34). As with the biotinylated species, small changes in the molar ratio of protein to conjugate resulted in a change in the distribution of molecular species as well as a change in their proportions. HEL has three tyrosine residues, and since each tyrosine has the potential to conjugate two iodine atoms, incorporation of up to six iodine atoms might be expected. However, it is well documented that the two most reactive tyrosine residues in HEL are Tyr-20 and Tyr-23. The third residue, Tyr-53, generally only reacts under extreme conditions (33). As indicated by the data in Table 1, excessive iodination of HEL only results in the incorporation of a maximum of four iodine atoms which is consistent with an unreactive third tyrosine

Bioconjugate Chem., Vol. 7, No. 1, 1996 21

Electrospray MS of Chemically-Modified Proteins

Table 2. Expected and Observed Masses for B72.3 Monoclonal Antibody and Fragments expected mass (Da)

obsd mass (Da)

46 83228,29

46 525 ( 8

25 739

25 733 ( 7

25 901

25 895 ( 6

26 063

26 057 ( 8

Fab fragment (light chain: Asp-1 f Cys-214 plus heavy chain fragment: Gln-1 f Gly-217) reduced Fc fragment (heavy chain fragment: Val-225 f Lys-438) + core carbohydratea + 2 N-acetylglucosamine reduced Fc fragment (heavy chain fragment: Val-225 f Lys-438) + core carbohydratea + 2 N-acetylglucosamine + 1 galactose reduced Fc fragment (heavy chain fragment: Val-225 f Lys-438) + core carbohydratea + 2 N-acetylglucosamine + 2 galactose B72.3 MAb a

143

75628,29 b

146 861 ( 52

b

Core carbohydrate ) (mannose)3(N-acetylgalactosamine)4(fucose). Molecular mass of nonglycosylated MAb.

residue. This would not be apparent from other methods used for assessment of iodine incorporation which only yield an average figure. Confirmation of a nonreactive Tyr-53 toward iodination would require further structural analysis by additional experiments involving enzymatic digestion and LC-MS or MS-MS analysis of the resulting peptides. The spectrum in Figure 4 indicates the addition of one [Co(diAMsar)]3+ moiety to HEL, despite the use of a 1000fold molar excess of ligand. A similar result was obtained by Conrad et al. (22) in coupling reactions of [Co(diAMsar)]3+ to cytochrome c. These workers also used more detailed mass spectrometric studies to demonstrate that there were a number of different attachment sites for this ligand on the protein; however, in all cases, only one [Co(diAMsar)]3+ was attached to the protein. In addition, the spectrum also shows that a number of N-acylurea EDC-modified glutamic or aspartic acid residues are also present but have not covalently bound the [Co(diAMsar)]3+ moiety. This would clearly be undesirable from the point of view of obtaining maximum incorporation of [Co(diAMsar)]3+. This type of structural detail could in turn be used in the optimization of the chemistry involved in conjugation of such ligands to the protein. ESI-MS can also provide an estimate of the degree of purity of the required compound. For example, if the desired product is the protein plus one ligand without EDC coupling groups and the ESI spectrum indicates that this is not the predominant component, the chemistry can be modified until this is shown to be the major component by ESI-MS. Alternatively, once it has been established by ESI-MS that the required species is present, ESI-MS can be used to monitor chromatographic separations to ensure maximum purity of the desired product. The peaks from the intact MAb (in Figure 5) are approximately 1300 Da (∼30 m/z units) wide due to both glycosylation, salt adducts, and the large number of isotopic states for a molecule of this size. Thus, it would not be possible to resolve a difference of say 226 Da from one biotin molecule which corresponds to ∼4 m/z units for the 50+ charge state and one galactose residue (162 Da, ∼3 m/z units). However by “oversmoothing” so that an average mass is determined it is likely that average mass shifts would be observed, as has been demonstrated with MALDI-MS (8). Alternatively, MAbs could be modified with the ligand of choice, enzymatically digested to Fab and Fc fragments, and the individual fragments assessed by ESI-MS to determine the loading values of the ligand on the fragments. We have already shown that the detection of a small ligand on a 50 kDa Fab fragment can readily be observed (Figure 8). Experiments are currently under way to remove the carbohydrate side chain from the CH2 region of B72.3. This will not only allow direct comparison of the molecular weight obtained by ESI-MS with the known DNA sequence of the antibody but it may enable chemical modifications

on a considerably larger protein to be observed. It should also be possible to use ESI-MS for assessing the purity of the MAb, i.e., to determine if one or more MAbs are present. The carbohydrate structure of murine IgG is an asparagine linked oligosaccharide of the biantennary complex type (35). However, instead of a single oligosaccharide, an array of structurally related yet distinct oligosaccharides is expressed. It has been reported that 94% of murine oligosaccharides have a fucosylated core and that 80% bear either no or a single galactose residue. Solely on the basis of mass, the proposed carbohydrate composition for the B72.3 MAb appears to have the same structure as that elucidated by Mizuochi et al. (35). The molecular mass of the B72.3 Fab fragment as determined from DNA sequencing is 46 832 Da assuming cleavage at the C-terminus of Gly-217 of the MAb heavy chain. This is 307 Da greater than the mass of the papain generated Fab fragment as determined by ESIMS. There are two plausible explanations for this observed discrepancy. Firstly, it is possible that there may be an error in the DNA sequence which has been translated to errors in the amino acid sequence, or secondly, it is feasible that the papain has overdigested the B72.3 MAb and removed amino acids from the Fab fragment. However, it is not obvious from the sequence and mass differences where this may have occurred. Such a difference may not be readily detected by conventional techniques. SDS-PAGE, for example, would be unable to discriminate between a mass of 46 832 and 46 525 Da. This clearly illustrates the power of ESI-MS for determining small changes in mass on large proteins. CONCLUSION

The work described here demonstrates the potential of electrospray mass spectrometry for the characterization of chemically-modified proteins including MAb fragments. ESI-MS offers the capability for determining both the numbers of ligands bound to the protein and the actual distributions of individual species, thus providing a degree of structural information not accessible by other methods. Thus, ESI-MS should prove to be a powerful aid in the design and development of improved radioimmunoconjugates for future clinical applications. Finally, we and others have demonstrated ESI-MS can be used for the measurement of intact MAbs thus providing a means of establishing the purity of antibody-based pharmaceuticals. ACKNOWLEDGMENT

K.L.B thanks the Australian Research Council (ARC) for an APRA (Industry) scholarship and the Australian Institute for Nuclear Science and Engineering (AINSE) for additional financial support. Grants from ARC, Ramaciotti Foundation, Ansto, and the University of

22 Bioconjugate Chem., Vol. 7, No. 1, 1996

Wollongong enabled the purchase of the electrospray mass spectrometer. LITERATURE CITED (1) Riethmu¨ller, G., Schneider-Ga¨dicke, E., and Johnson, J. P. (1993) Monoclonal antibodies in cancer therapy. Curr. Opin. Immunol. 5, 732-739. (2) James, K. (1993) Human monoclonal antibodies and engineered antibodies in the management of cancer. Cancer Biol. 1, 243-253. (3) Hiltunen, J. V. (1993) Search for new and improved radiolabelling methods for monoclonal antibodies. Acta Oncol. 32, 831-839. (4) Griffiths, G. L., Goldenberg, D. M., Jones, A. L., and Hansen, H. J. (1992) Radiolabelling of monoclonal antibodies and fragments with technetium and rhenium. Bioconjugate Chem. 3, 91-99. (5) Khazaeli, M. B., Conry, R. M., and LoBuglio, A. F. (1994) Human immune response to monoclonal antibodies. J. Immunother. 15, 42-52. (6) Pietersz, G. A., Cunningham, Z., and McKenzie, I. F. C. (1988) Specific in vitro anti-tumour activity of methotrexatemonoclonal antibody conjugates prepared using human serum albumin as an intermediary. Immunol. Cell Biol. 66(1), 4349. (7) Meares, C. F., McCall, M. J., Reardan, D. T., Goodwin, D. A., Diamanti, C. I., and McTigue, M. (1984) Conjugation of antibodies with bifunctional chelating agents: isothiocyanate and bromoacetamide reagents, methods of analysis and subsequent addition of metal ions. Anal. Biochem. 142, 6878. (8) Siegel, M. M., Hollander, I. J., Hamann, P. R., James, J. P., Hinman, L., Smith, B. J., Farnsworth, A. P. H., Phipps, A., King, D. J., Karas, M., Ingendoh, A., and Hillenkamp, F. (1991) Matrix-assisted UV-laser desorption/ionisation mass spectrometric analysis of monoclonal antibodies for the determination of carbohydrate, conjugated chelator and conjugated drug content. Anal. Chem. 63, 2470-2481. (9) Fenn, J. B., Mann, M., Meng, C. K., Wong, S. F., and Whitehouse, C. M. (1989) Electrospray ionisation for mass spectrometry of large biomolecules. Science 246, 64-71. (10) Smith, R. D., Loo, J. A., Edmonds, C. G., Barinaga, C. J., and Udseth, H. R. (1990) New developments in biochemical mass spectrometry: electrospray ionisation. Anal. Chem. 62, 882-899. (11) Feng, R., and Konishi, Y. (1992) Analysis of antibodies and other large glycoproteins in the mass range of 150 000200 000 Da by electrospray ionisation mass spectrometry. Anal. Chem. 64, 2090-2095. (12) Feng, R., Konishi, Y., and Bell, A. W. (1991) High accuracy molecular weight determination and variation characterisation of proteins up to 80 ku by ionspray mass spectrometry. J. Am. Soc. Mass Spectrom. 2, 387-401. (13) Jiskoot, W., van de Werken, G., Coco Martin, J. M., Green, B. N., Beuvery, E. C., and Crommelin, D. J. A. (1992) Application of electrospray mass spectrometry (ES-MS) for the analysis of monoclonal antibody Fc subunits. Pharm. Res. 9, 945-951. (14) Feng, R., and Konishi, Y. (1993) Collisionally-activated dissociation of multiply charged 150-kDa antibody ions. Anal. Chem. 65, 645-649. (15) Lewis, D. A., Guzzetta, A. W., Hancock, W. S., and Costello, M. (1994) Characterization of humanised anti-TAC, an antibody directed against the interleukin 2 receptor, using electrospray ionisation mass spectrometry by direct infusion, LC/MS, and MS/MS. Anal. Chem. 66, 585-595. (16) Verentchikov, A. N., Ens, W., and Standing, K. G. (1994) Reflecting time-of-flight mass spectrometer with an electrospray ion source and orthogonal extraction. Anal. Chem. 66, 126-133.

Bennett et al. (17) Bourell, J. H., Clauser, K. P., Kelley, R., Carter, P., and Stults, J. T. (1994) Electrospray ionisation mass spectrometry of recombinantly engineered antibody fragments. Anal. Chem. 66, 2088-2095. (18) Ashton, D. S., Beddell, C. R., Cooper, D. J., Craig, S. J., Lines, A. C., Oliver, R. W. A., and Smith M. A. (1995) Mass spectrometry of the humanised monoclonal antibody CAMPATH 1H. Anal. Chem. 67, 835-842. (19) Colcher, D., Esteban, J. M., Carrasquillo, J. A., Sugarbaker, P., Reynolds, J. C., Bryant, G., Larson, S. M., and Schlom, J. (1987) Quantitative analyses of selective radiolabelled monoclonal antibody localisation in metastatic lesions of colorectal cancer patients. Cancer Res. 47, 1185-1189. (20) Bayer, E. A., and Wilchek, M. (1990) Protein Biotinylation. Methods in Enzymology Vol. 184, pp 138-157, Academic Press, San Diego, CA. (21) Fraker, P. J., and Speck, J. C., Jr. (1978) Protein and cell membrane iodination with a sparingly soluble chloroamide, 1,3,4,6-tetrachloro-3R,6R-diphenylglycoluril. Biochem. Biophys. Res. Commun. 80, 849-857. (22) Conrad, D. W., Zhang, H., Stewart, D. E., and Scott, R. A. (1992) Distance dependence of long-range electron transfer in cytochrome c derivatives containing covalently attached cobalt cage complexes. J. Am. Chem. Soc. 114, 9909-9915. (23) Geue, R. J., Hambley, T. W., Harrowfield, J. M., Sargeson, A. M., and Snow, M. R. (1984) Metal ion encapsulation: cobalt cages derived from polyamines, formaldehyde and nitromethane. J. Am. Chem. Soc. 106, 5478-5488. (24) Ferrige, A. G., Seddon, M. J., Green, B. N., Jarvis, S. A., and Skilling, J. (1992) Disentangling electrospray spectra with maximum entropy. Rapid Commun. Mass Spectrom. 6, 707-711. (25) Kilby, G. W., Truscott, R. J. W., and Sheil, M. M. Unpublished observations. (26) Mather, S. J. (1986) Radioiodinated monoclonal antibodies: a critical review. Appl. Radiat. Isot. 37, 727-733. (27) Parker, D. (1990) Tumour targetting with radiolabelled macrocycle-antibody conjugates. Chem. Soc. Rev. 19, 271291. (28) Kabat, E. A., Wu, T. T., Reid-Miller, M., Pery, H. M., and Gottesman, K. S. (1987) Sequences of Proteins of Immunological Interest, 4th ed., U.S. Department of Health and Human Services, National Institute of Health, Bethesda, MD. (29) Whittle, N., Adair, J., Lloyd, C., Jenkins, L., Devine, J., Schlom, J., Raubitschek, A., Colcher, D., and Bodmer, M. (1987) Expression in COS cells of a mouse-human chimaeric B72.3 antibody. Protein Eng. 1, 499-505. (30) Goding, J. W. (1986) Monoclonal Antibodies: Principles and Practice, 2nd ed., Academic Press, London, UK. (31) Bennett, K. L., Hick, L. A., Truscott, R. J. W., Sheil, M. M., and Smith, S. V. (1995) Optimum conditions for electrospray mass spectrometry of a monoclonal antibody. J. Mass Spectrom. 30, 769-771. (32) Rothman, R. J., Warren, L., Vliegenthart, J. F. G., and Ha¨rd, K. J. (1989) Clonal analysis of the glycosylation of immunoglobulin G secreted by murine hybridomas. Biochemistry 28, 1377-1384. (33) Imoto, T., Johnson, L. N., North, A. C. T., Phillips, D. C., and Rupley, J. A. (1972) Vertebrate Lysozymes. The Enzymes, 3rd ed., (P. D. Boyer, Ed.) Vol. 7 pp 665-868, Academic Press, New York. (34) Seevers, R. H., and Counsell, R. E. (1982) Radioiodination techniques for small organic molecules. Chem. Rev. 82, 575590. (35) Mizuochi, T., Hamako, J., and Titani, K. (1987) Structures of the sugar chains of mouse immunoglobulin. Arch. Biochem. Biophys. 257, 387-394.

BC950064C